My Tenure as the Editor-in-Chief of Computational Linguistics

The field of Computational Linguistics and Natural Language Processing has undergone rapid changes. Authors have come to expect a rapid publication cycle. Before my time, the journal was run by an editorial board, where editorial board members served as the reviewers, augmented by additional reviewers recruited by the editor-in-chief for reviewing specific papers. In order to speed up the reviewing process, I introduced a two-tier reviewing structure, similar to the reviewing practice at conferences. The CL editorial board now comprises action editors, who oversee the reviewing of submitted papers. I assign each submitted paper to an action editor, who is then in charge of inviting three reviewers to review the paper. The action editor also provides a meta-review of the paper, and recommends whether the paper should be accepted, revised, or rejected. The name of the action editor who oversees the reviewing of a paper and recommends its acceptance is listed on each published CL journal paper.

In addition, in order to provide a quick turnaround time and to provide a steady pool of reviewers, I have recruited a team of standing reviewers at CL. A standing reviewer agrees to review four submitted papers in a year and to return the review within three weeks for a paper.

I am pleased to report that with these changes in the CL workflow, the average time to first decision for regular papers and full survey papers (excluding desk rejects) is 47 days in 2022 and 46 days for the first ten months of 2023 (1 January to 30 October 2023). This is a respectable turnaround time, even shorter than the current two-month reviewing cycle for ACL Rolling Review (ARR). Note that this turnaround time is achieved while still allowing authors to submit their longer journal papers to CL at any time, without waiting for the bimonthly submission deadline like in ARR.

One attractive aspect of publishing conference papers is that authors get to present their papers at a conference, which gives them an opportunity to present their work before their peers and to receive feedback and comments at the conference. In order to provide this attractive feature of a conference paper, the authors of an accepted Computational Linguistics regular paper (which is not an extension of a prior published paper that has been presented at a previous conference or workshop) can opt to present their accepted paper at one of the annual conferences in the ACL family (ACL, AACL, EACL, NAACL, and EMNLP). Since presentation of accepted CL papers was started at the ACL 2020 conference, authors of almost all accepted CL papers have opted to present their papers at a conference, confirming the popularity of the conference presentation option.

The standard and widely used measure of a journal’s influence is its impact factor, compiled by Journal Citation Reports (JCR). Figure 1 shows the impact factor of Computational Linguistics over the years since its impact factor was first tracked in 1997.

For a given year, the impact factor is the ratio between (1) the number of citations received in that year for publications in that journal that were published in the two preceding years, and (2) the total number of publications in that journal during the two preceding years. For example, the impact factor of CL in 2021 is calculated based on the citations received in 2021 to CL papers published in 2019 and 2020.

The impact factor of Computational Linguistics has historically been lower than 3.0 before 2021. I am pleased to report that CL’s impact factor has risen sharply to 7.778 and 9.3 in 2021 and 2022, respectively, the most recent two years that impact factors were available.

The number of submissions to CL (counting regular papers, survey proposals, and full survey papers) per year averages about 140 over 2018 – 2022. However, the number of submissions has risen sharply from 138 in 2022 to 215 in 2023, representing a 55% increase in number of submissions.

In addition, there is a healthy pipeline of accepted papers at the journal. For example, at the time of writing (December 2023), there are eleven accepted regular papers, two conditionally accepted regular papers, two accepted survey papers, and one conditionally accepted survey paper.

Moreover, the call for papers for a special issue of Computational Linguistics on Language Learning, Representation, and Processing in Humans and Machines has attracted 19 submissions. The guest editors of this special issue are Marianna Apidianaki, Abdellah Fourtassi, and Sebastian Padó, and the special issue is slated for publication in the December 2024 issue of CL.

Computational Linguistics also faces challenges. Researchers in CL and NLP now favor rapid dissemination of their research findings. They release their papers on arXiv, and expect their papers to undergo rapid review and publication. On the other hand, it is becoming increasingly challenging to find reviewers who are willing to accept to review papers to meet the rapid turnaround time expected by authors. Even with the introduction of standing reviewers at CL, there are standing reviewers who decline their reviewing assignment due to various reasons. So some delay is introduced due to the time taken to find three reviewers who agree to review a paper, and the extra time taken by reviewers who do not submit their reviews on time.

In addition, there are also multiple publication venues available to CL and NLP authors. Besides the annual conferences in the ACL family, there is also another journal TACL (Transactions of the Association for Computational Linguistics) to which they can submit their papers.

While a journal paper is typically longer than a conference paper and so a journal caters to publishing work in a longer form, this situation is also changing. Currently, Computational Linguistics only requires a submitted paper to be at least 15 pages long, so short journal papers have always been allowed at Computational Linguistics. Traditionally, a journal can publish a longer extended version of a conference paper, and CL still accepts such journal submissions. However, conferences in the ACL family now allow a conference paper to have an unlimited number of pages for appendices and supplementary material. Hence, the need for publishing an extended version of a conference paper also diminishes.

In closing, I would like to thank the editorial team at Computational Linguistics for their dedicated service. They include the book review editor Aline Villavicencio, the squibs editor Michael White, the editorial assistant Muhammad Reza Qorib, the action editors, the standing reviewers, and all additional reviewers recruited to review papers for the journal. I would also like to thank all authors who choose to submit their papers to Computational Linguistics.